MCP-Persona: Benchmarking LLM Agents on Real-World Personal Applications via Environment Simulation

Researchers introduced MCP-Persona, the first benchmark for evaluating LLM agents on personalized MCP tools in simulated environments.

The benchmark includes real-world apps such as Reddit, Xiaohongshu, Lark, and Slack, aiming to test more realistic agent behavior.

According to the authors, current state-of-the-art agents still struggle on these personalized tasks.

The work highlights gaps that generic tool benchmarks often miss and offers a new way to measure agent performance.